(Received 15 Sep. 2009: accepted 1 Feb. 2010)

†Fax: 018-872-1676，E-mail: Jiey_chen@akita-pu.ac.jp

1．Introduction

Potatoes are used in the food industry to produce food

products such as potato flakes, potato chips, and French

fries. In each instance, potatoes are presorted before

being introduced to the processing line. Consequently,

the composition of the potato must be measured for vari-

ables such as dry matter, carbohydrate, and reducing

sugar content (fructose and glucose) prior to determin-

ing its correct use, handling, and processing [1-3]. In

the fried products industry, the determination of the

reducing sugars content is par ticularly impor tant

because the presence of high levels of reducing sugars

not only causes browning, but also generates acrylic

amide by the Maillard reaction with asparagines during

the high temperature heating process. Acrylic amide is

believed to be carcinogenic and poisonous to the ner-

vous system and is therefore a highly undesirable attri-

bute for consumers.

Potato tubers are normally stored at low temperatures

to prevent sprouting and weight loss; however, during

storage sugar content increases. Freshly har vested

tubers normally have sugars in trace amounts. Very high

sugar levels may be reached when stored at low tempera-

tures. Removal of these potatoes with high reducing

sugar content is necessary in the fried food industry.

The reducing sugar content of potatoes is generally

determined by very accurate chemical methods, such as

high performance liquid chromatography (HPLC). These

methods are performed off-line, are time consuming and

usually require trained personnel for their execution.

Additionally, these methods destroy the material that is

being analyzed. Due to these reasons, chemical methods

are generally not feasible for on-line measurement appli-

cations in the food industry. Therefore, a rapid method for

determining the amounts of sugars in intact tubers would

be useful for monitoring sugar levels during extended

storage or for testing of shipments prior to processing.

Near infrared (NIR) spectroscopy offers a rapid and

nondestructive alternative to chemical measurement

techniques to evaluate compounds in food. NIR spectros-

copy has been used to nondestructively estimate the spe-

cific gravity of potato tubers [4] and dry matter in pota-

toes [5], as well as the main ingredients including the

carbohydrate content in potatoes [6-9]. In contrast, there

are no reports describing nondestructive measurement

of the sugar content in intact potatoes; however, a

destructive NIR analysis of the sugar content of potato

slices was reported [10,11]. The objective of this study

was to ascertain whether NIR spectroscopy could be

used to determine the reducing sugar content of intact

potato tubers.

2．Materials and methods

2.1 Experimental sample preparation

Potatoes of one cultivar (May-Queen) were purchased

Nondestructive Determination of Sugar Content in Potato Tubers Using Visible and Near Infrared Spectroscopy

Jie Yu CHEN1,†, Han ZHANG1, Yelian MIAO2, and Mitsunaka ASAKURA1

1Faculty of Bioresource Sciences, Akita Prefectural University 241-438 Kaidobata-Nishi, Shimoshinjo-Nakano Akita-shi, Akita 010-0195, Japan

2College of Food and Light Industrial Engineering, Nanjing University of Technology, Nanjing 210009, China

Near infrared (NIR) spectroscopy was investigated as a method for the nondestr uctive measurement of sugar content in intact potato tubers. The NIR spectra (400-1100 nm) of potato samples were acquired by fiber optics in the interactance mode. Calibration models for the prediction of the fructose and glucose contents were developed by a partial least squares statistical analysis method. The calibration model gave standard errors of prediction of 0.26 mg/g for fructose and 0.46 mg/g for glucose. The preliminary results showed that NIR spectroscopy gives a reasonable estimate that can be used for nondestructive sorting of potatoes according to their sugar content.Key words: nondestructive determination, sugar, potato, near infrared spectroscopy, interactance

◇◇◇ Original Paper ◇◇◇

Japan Journal of Food Engineering, Vol. 11, No. 1, pp. 59 - 64, Mar. 2010

Jie Yu CHEN, Han ZHANG, Yelian MIAO and Mitsunaka ASAKURA60

on the 7th of September, 2008 from the Akita fruits and

vegetables distribution market and were used as the

experimental samples. The potato samples were grown at

locations around Hokkaido, Japan. A total of 100 potato

samples were used. Potato samples were individually

separated into two groups. To provide a wide range of

sugar concentrations in the sample groups, one group

was stored at 5℃ for 4 months, and the other group was

stored at 25℃ for 4 months. At biweekly intervals, five

tubers were removed from each storage room for testing.

2.2 Spectra acquisition

NIR spectra were measured on all intact potato sam-

ples using a spectrophotometer (NIRSystems Model

6500) equipped with a fiber optics interactance probe as

shown in Fig. 1. The fiber optic probe consisted of a cen-

tral bundle (diameter: 7.6 mm) of Schott glass fibers and

a concentric ring of the fibers (outside diameter: 19 mm,

width: 0.64mm). The ring was separated from the central

bundle by a 5 mm thick metal barrier. To obtain the opti-

cal spectrum, an intact potato is placed on the optical

probe and the cover of the sample-housing is closed.

Monochromatic light is emitted by the ring, and "inter-

acts" with the tissue. Some of the non-absorbed light is

internally reflected to the central bundle. NIR spectra in

the wavelength range of 400-1100 nm at 2 nm intervals

were collected as absorbance measurements. A 2.5-mm

thick white ceramic plate was used as a standard optical

reference. Each spectrum was an average of 32 scans

and one spectrum was obtained per potato sample. Prior

to spectral measurements, each sample was placed in the

measurement room (25℃) for 5 hours.

Raw spectra obtained from three differently sized pota-

toes are shown in Fig. 2. The spectra clearly shift accord-

ing to the sample size. To reduce the sample-size effect,

we used the second derivative spectra. In addition, the

second derivative spectra in the 600–1100 nm wavelength

range showed higher resolution than the spectra in the

400–600 nm wavelength range. Thus, second derivative

spectra were used to estimate sugar content of stored

potatoes in this study.

2.3 Chemical analysis

After NIR spectra were measured, flesh of the potato

tuber samples was cut and mashed by a mixer, the mate-

rial was wrapped in gauze, and then the juice was

extracted by squeezing. This juice was used for sugar

analysis. The sugar compositions (fructose, glucose, sor-

bitol, sucrose) of the potato juices were analyzed using

an HPLC (Shimadzu, SCL-10A VP) method with a refrac-

tive index detector (Shimadzu, RID-10A). Sample mix-

tures were separated on a Shodex Asahipak NH2P-50 4A

column using an acetonitrile/water (3:1 v/v) solution as

the eluate. The operating conditions were as follows: flow

rate of 1 mL/min; column temperature of 40℃; injection

sample volume was 10μL. The raw potato juice samples

were centrifuged at 3000 rpm for 30 min, diluted 1:5 by

adding distilled water and subsequently passed through

a 0.45-μm pore-size filter.

2.4 Statistical Analysis

Data analyses were carried out using "Unscrambler"

(CAMO, Oslo, Norway) and Excel software (Microsoft

Office 2003). Second derivative spectra (Savitzky-Golay

algorithm, left and right averaging of 20 nm and 2nd

order polynomial) were used. The partial least square

(PLS) regression was used to develop calibrations for

analyzing the sugar compositions of intact potato tuber

samples. Calibrations were performed on glucose and

fructose. PLS regression was performed in the wave-

Fig. 1 The configuration of the interactance probe. Fig. 2 Raw visible and NIR spectra of three dif ferently sized tubers.

Potato

From light source

To detector

A A

A-A To detector

From light source

Determination of Sugar Content in Potato Tubers 61

length range of 600 to 1100 nm where the absorption

bands were assigned to the fourth, third, and second

overtones of O-H and C-H fundamental bands or their

combinations. Validation was performed by leave-one-

out full cross validation.

3．Results and discussion

3.1 Variations in the sugar content of potato

tubers during storage

Figure 3 shows the variations in the contents of glu-

cose and fructose of potato tubers stored at either 5 or

25℃, as measured by chemical methods. The sugar con-

tent values shown are averages of five potato tubers. In

tubers stored at 5℃, the glucose contents of the potato

tuber samples gradually increased after 1 week of stor-

age and continued to increase until 8 weeks of storage.

The sugar contents of the potato tuber samples

decreased after the 8th week of storage, suggesting that

the potato tubers had begun to germinate. Fructose lev-

els in the potato tuber samples also increased until the

8th week of storage. For potato tubers stored at 25℃,

there were no clear variations in the glucose and fruc-

tose contents at the beginning of storage, but clear

decreases were observed after storage for 6 weeks.

3.2 Variations of the NIR spectra during the

storage process

The average NIR second derivative spectra of 15 potato

tuber samples stored at 25℃ for 0, 2, 4, 6, 8, 10, 12, 14,

and 16 weeks are shown in Fig. 4(a). The variations in

the second derivative spectra that result from the chang-

ing composition of the stored tubers may be very subtle,

and therefore, can be difficult to observe. To clearly

observe and confirm the variations in spectral intensity,

the variance of these average spectra was calculated and

plotted as shown in Fig. 4(b). In the variance plot of the

second derivative spectra, the peaks around 970 nm and

760 nm were attributable to water. The differences in the

sizes of these peaks indicated that the potato tuber sam-

ples lost weight during the storage process. Other peaks

were observed around 670, 710, 798, 830, and 926 nm.

The peak at 670 nm is attributable to chlorophyll, which

is present in greater quantities in immature fruit than in

mature fruit [12]. However, we also observed an absorp-

tion peak of glucose and fructose around 670 nm in the

second derivative spectra of glucose and fructose (Fig.

5). Moreover, the NIR second derivative spectra of potato

tuber samples stored at 5℃ showed relatively large varia-

tions around 670 and 710 nm (Fig. 6(b)). Based on these

facts, the considerable variations around 670 and 710 nm

may be associated, either directly or indirectly, with the

sugar content of potato tubers (see Fig. 4(b) for compari-

Fig. 3 Changes in the amounts of reducing sugars in the potato tubers stored at 5℃ (a) or 25℃ (b).

Fig. 4 Variations in NIR second derivative spectra of potato tubers during storage at 25℃ for 16 weeks (a), and variance of NIR second derivative spectra (b).

0.00

0.50

1.00

1.50

2.00

2.50

0 2 4 6 8 10 12 14 16

Storage time (weeks)

Sug

ar c

onte

nt (m

g/g) Fructose

Glucose

0.00

0.50

1.00

1.50

2.00

2.50

0 2 4 6 8 10 12 14 16

Storage time (weeks)

Sug

ar c

onte

nt (m

g/g) Fructose

Glucose

(a)

(b)

Jie Yu CHEN, Han ZHANG, Yelian MIAO and Mitsunaka ASAKURA62

son). The spectra values at 710 nm and 934 nm are the

fourth and third overtones of C-H stretching [13]. The

values at 798 nm and 826 nm also may be related to fruc-

tose and glucose (Fig. 5).

3.3 PLS regression

PLS regression analysis was performed based on the

sugar (glucose and fructose) contents and NIR second

derivative spectra (600-1100 nm) of all potato tuber sam-

ples (total of 80: 40 stored at 5℃ and 40 stored at 25℃).

Figure 7 shows the calibration and validation results of

the fructose (a) and glucose (b) contents of the potato

tubers. The correlation coefficients of the calibration

were 0.71 and 0.65, respectively, with a standard error of

validation (SECV) equal to 0.26 mg/g and 0.46 mg/g,

respectively. As an index for determining the validity of

the calibration models, the RPD (ratio of standard devia-

tion of reference data in prediction sample set to SECV)

is usually employed [14]. A RPD value of 1.4-1.7 is

regarded as adequate for rough screening and a value

above 1.7 is regarded as satisfactory for screening. In this

study, the RPD values of 1.42 for fructose content and

1.32 for glucose (close to the value proposed by Williams)

were obtained. As such, the NIR calibration models are

suitable for rough screening of the sugar contents of the

potato tubers. Here, this nondestructive sorting approach

can roughly partition the potato tubers into two groups;

those with high values, and those with low values.

3.4 Discussion of the NIR Calibration Models

Regression coefficients can be used to discuss the con-

tributions of individual wavelengths to a PLS calibration

model, because a regression coefficient spectrum shows

characteristic peaks and troughs that can indicate which

wavelength range is important for the calibration model

[15,16]. Figure 8 shows the regression coefficients of the

PLS calibration models of the fructose (a) and glucose

(b) components, respectively. These regression coeffi-

cient spectra showed several marked peaks, such as

those observed at 670, 710, 798, and 830 nm. Variations

in the size of these peaks of the NIR second derivative

spectra were observed in potato tuber samples during

storage. We confirmed that these peaks were the absorp-

tion bands of glucose or fructose by comparison to sec-

ond derivative spectra of glucose and fructose standards

(Fig. 5). In particular, the peak around 798 nm is consis-

tent with the findings of Miyamoto and Kitano [17], who

reported a sugar absorbance band around 794 nm. The

results presented here are also consisted with those

reported by McGlone [18] and Abebe [19], who reported

a calibration model to estimate soluble solids content of

mandarin fruit. In addition, there is a small peak around

830 nm in the regression coefficient of the PLS calibra-

tion model that can only be accounted for by glucose, as

shown in Fig. 8. We confirmed that this peak represented

the absorption band of glucose by comparison to the sec-

ond derivative spectra of glucose and fructose standards

(Fig. 5). In other words, it is possible to differentiate glu-

cose from fructose on the basis of these small differences

in the calibration models. Furthermore, we confirmed

that the peak at 888 nm in the regression coefficient

Fig. 5 Second derivative spectra of standard chemical reagents of glucose and fructose.

Fig. 6 Variations in NIR second derivative spectra of potato tubers during storage at 5℃ for 16 weeks (a) and variance of these NIR second derivative spectra (b).

-0.006-0.005-0.004-0.003-0.002-0.001

00.0010.0020.0030.0040.005

600 700 800 900 1000 1100

Wavelength (nm)

log(

1/R

)

670

710

798

830888

918992

glucose

fructose

Determination of Sugar Content in Potato Tubers 63

spectra represented the absorption band of fructose, as

shown in Fig. 5. Consequently, the PLS calibration mod-

els for estimating the sugar components of potatoes were

established based on the specific absorption characteris-

tics of the various sugar components.

3.5 NIR Prediction Results

The PLS calibration models for estimating the sugar

components were applied to other independent potato

tuber samples. The potato samples were separated into

two groups; one was stored at 5℃ for 3 months, the other

was stored at 25℃ for 3 months. As in the calibration

analysis, five tubers were removed from each storage

room every 2 weeks, and their NIR spectra were evalu-

ated. Fructose and glucose contents of the potato tubers

were predicted based on the calibration models. The

changes in the glucose and fructose levels of the potato

tuber samples during storage are shown in Fig. 9. These

results were consistent with pattern of glucose and fruc-

tose values determined in the potato tuber samples using

chemical methods.

4．CONCLUSION

In this study, we roughly estimated reducing sugar

contents in potato tubers during storage using a simple

non-destructive approach that involved the acquisition of

NIR spectra (400-1100 nm) with fiber optics in the inter-

actance mode. The marked variations around 670, 798,

830, and 888 nm were associated with sugar content in

the potato tubers. Comparisons between variance plots of

the second derivative NIR spectra of potato samples

stored at 25 and 5℃ showed clear differences that were

attributable to sugar and water content. Based on the

variations around these major wavelength bands, calibra-

tion models to accurately estimate fructose and glucose

contents were developed by a partial least squares statis-

tical analysis method. This preliminary study demon-

strates that near infrared spectroscopy can be used for

Fig. 7 Scatter plots of sugar contents of potato tubers, and values predicted using NIR technique.

-1500

-1200

-900

-600

-300

0

300

600

900

600 700 800 900 1000 1100

Wavelength (nm)

Reg

ress

ion

coef

ficie

nts

710

(a) Fructose

-2500

-2000

-1500

-1000

-500

0

500

1000

1500

600 700 800 900 1000 1100

Wavelength (nm)

Reg

ress

ion

coef

ficie

nts

710

(b) Glucose

670

670

888

888

798

800

830

Fig. 8 Regression coefficients of PLS calibration models based on second derivative NIR spectra (600-1100 nm) for glucose (a) and fructose (b) contents of potato tubers.

0.0

0.3

0.6

0.9

1.2

1.5

1.8

0.0 0.3 0.6 0.9 1.2 1.5 1.8Fructose content (mg/g)

NIR

pre

dict

ed v

alue

(mg/

g)

R=0.71SEC=0.23 mg/gSECV=0.26 mg/g

0.0

0.4

0.8

1.2

1.6

2.0

2.4

2.8

0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8Glucose content (mg/g)

NIR

pre

dict

ed v

alue

(mg/

g)

R=0.65SEC=0.39 mg/gSECV=0.46 mg/g

(a) Fructose (b) Glucose

Jie Yu CHEN, Han ZHANG, Yelian MIAO and Mitsunaka ASAKURA64

nondestructive sorting of potatoes according to their

sugar content. However, the accuracy of this method

could still be improved, and further research is required

to develop the method to a level that is suitable for practi-

cal use.

REFERENCES

1) Y. Umemura; The production, circulation and processing of

agricultural products for meeting the consumption needs

(in Japanese). Presentation at the 50th Japanese Society of

Agricultural Machinery Hokkaido Branch meeting, 76, p.73

(1999).

2) G. Lisinska; Manufacture of potato chips and French fries.

“Potato Science and Technology”, G. Lisinska and W.

Leszczynski ed., Elsevier, Barking, 1989, pp.165-232.

3) O. Smith, C. O. Davis; Production, Storing, Processing in

potatoes. “Potato processing”, 2nd ed., O. Smith ed, Avi

Publishing Company, Westport, CT, 1977, pp.677-724.

4) J. Y. Chen, H. Zhang, Y. Miao, R. Matsunaga; NIR measure-

ment of specific gravity of potato. Food Sci. Technol. Res.,

11, 26-31 (2005).

5) G. G. Dull, G. S. Birth, R. G. Leffler; Use of near infrared

analysis for the nondestructive measurement of dry matter

in potatoes. Am. Potato J., 66, 215-225 (1989).

6) J. Y. Chen, Y. Miao, H. Zhang, R. Matsunaga; Non-destructive

determination of carbohydrate content in potatoes using

near infrared spectroscopy. J. Near infrared Spectrosc., 12,

311-314 (2004).

7) S. Komiyama, J. Kato, H. Honda, K. Matsushima;

Development of sorting system based on potato starch con-

tent using visible and near infrared spectroscopy. Nippon

Shokuhin Kagaku Kogaku Kaishi, 54, 304-309 (2007).

8) N. U. Haase; Rapid estimation of potato tuber quality by near

infrared spectroscopy. Starch/Starke, 58, 268-273 (2006).

9) J. Y. Chen, H. Zhang, Y. Yagi; Nondestructive determination

of major components and pasting viscosity of potato tuber

by near infrared spectroscopy. Nippon Shokuhin Kagaku

Kogaku Kaishi, 56, 299-306 (2009).

10) M. Mehrubeoglu, G. L. Cote; Determination of total reducing

sugars in potato samples using near infrared spectroscopy.

Cereal Foods World, 42, 409-413 (1997).

11) M. G. Scanlon, M. K. Pritchard, L. R. Adam; Quality evalu-

ation of processing potatoes by near infrared reflectance. J.

Sci. Food Agric. 79, 763-771 (1999).

12) D. C. Slaughter; Nondestructive determination of internal

quality in peaches and nectarines. Trans. of the ASABE. 38,

617-623 (1995).

13) B. G. Osborne, T. Fearn, P.H. Hindle; Practical NIR spectros-

copy: with applications in food and beverage analysis. John

Wiley & Sons, Inc., 605 Third Avenue, New York, 1993, pp29.

14) P. C. Williams; Near-infrared technology in the agricultural

and food industries, and edition, Chapter 8, Implementation

of near-infrared technology, edited by P. Williams and K.

Norris, American Assoc. of Cereal Chemists Inc., Minnesota,

USA, 2001, pp145.

15) H. Martens, T. Naes; Multivariate calibration. John Wiley and

Sons: Chichester, UK, 1989, pp155.

16) J. Y. Chen, H. Zhang, R. Matsunaga; Rapid determination of

the main organic acid composition of raw Japanese apricot

fruit juices using near-infrared spectroscopy. J. Agric. Food

Chem., 54, 9652-9657 (2006).

17) K. Miyamoto, Y. Kitano; Non-destructive determination of

sugar content in Satsuma mandarin fruit by near infrared

transmittance spectroscopy. J. Near infrared Spectrosc. 3,

227-237 (1995).

18) V. A. McGlone, D. G. Fraser, R. B. Jordan,R. Kunnemeyer;

Internal quality assessment of mandarin fruit by vis/NIR

spectroscopy. J. Near infrared Spectrosc. 11, 323-332 (2003).

19) A. T. Abebe; Total sugar and maturity evaluation of intact

watermelon using near infrared spectroscopy. J. Near infra-

red Spectrosc. 14, 67-70 (2006).

Fig. 9 Variations in reducing sugar contents predicted by NIR technique for potato tubers stored at 5℃ (a) or 25℃ (b).

0.00

0.50

1.00

1.50

2.00

2.50

0 2 4 6 8 10 12 14

Storage time (weeks)

Sug

ar c

onte

nt (m

g/g) Fructose

Glucose

0.00

0.50

1.00

1.50

2.00

2.50

0 2 4 6 8 10 12 14

Storage time (weeks)

Sug

ar c

onte

nt (m

g/g) Fructose

Glucose

(a)

(b)

可視および近赤外分光法によるじゃがいもの糖成分の非破壊測定

陳　介余 1,†，張　函1，繆　冶煉2，朝倉　弘仲1

1秋田県立大学生物資源科学部，2中国南京工業大学食品与軽工学院

じゃがいもの糖成分の非破壊測定法の開発のため，可視および近赤外分光法の利用を検討した．光ファイバを利用したインタラクタンス法でじゃがいもの可視および近赤外スペクトル（400-1100 nm）を非破壊的に測定できた．測定されたスペクトルと糖成分の関係を検討した上で，PLS回帰分析法を用いて糖含量の予測モデルを開

発したところ，2次微分スペクトルとフルクトースおよびグルコース含量の間に有意の相関関係が認められ，標準誤差がそれぞれ 0.26 mg/gと 0.46 mg/gの予測モデルが得られた．本研究では，インタラクタンス測定法を用いた近赤外分光法がじゃがいもの糖含量の非破壊的測定法として利用の可能性があることを示唆した．

（受付 2009年 9月 15日，受理 2010年 2月 1日）

1 〒 010-0195　秋田市下新城中野字街道端西 241-438

2 〒 210009　中国南京市新模範馬路 5号

†Fax: 018-872-1676，E-mail: Jiey_chen@akita-pu.ac.jp

「日本食品工学会誌」, Vol. 11, No. 1, p. 65, Mar. 2010

◇◇◇ 和文要約 ◇◇◇